The minus of a plus is a minus. Mass death of selected neuron populations in sporadic late-onset neurodegenerative disease may be due to a combination of subtly decreased capacity to repair oxidative DNA damage and increased propensity for damage-related apoptosis

ABSTRACT

Neurons in the adult central nervous system (CNS) are subjected to high levels of oxidative damage that is usually promptly repaired. Transcribed genomic regions are repaired with priority over untranscribed regions. The prioritization of DNA repair in neurons results in modification of the input into the assessment of genomic integrity in order to delay or avoid damage-related apoptosis unless the damage interferes directly with the functioning of the neuron. CNS neurons may be replaced, albeit rarely. Over-stimulation of adult neural progenitor niche caused by accelerated neuronal loss may result in its premature depletion. The combination of the two pathologic mechanisms (increased rates of neuronal death and depletion of the progenitor niche) may eventually result in irreversible loss of specific cell populations in the CNS and/or generalized neuronal loss. Here we propose that the risk of developing sporadic late-onset neurodegenerative disease (LONDD) may be modulated by the individual capacity for detection and repair of DNA damage and the genetic propensity to repair moderate-degree damage or to assess it as irreparable and route the cell towards apoptosis. Thus, subtly deficient DNA damage repair coupled with a tendency to repair the damage rather than kill the damaged cell may be associated with increased risk of cancer, whereas deficient DNA repair coupled with a propensity to destroy damaged cells may increase the risk of LONDD. Extensive studies of individual repair capacity may be needed to test this hypothesis and, potentially, use the results in the assessment of the risk of common late-onset disease.

KEYWORDS:

• Late-onset disease
• neurodegeneration
• individual repair capacity
• assessment of genomic integrity

Abbreviations

8-OHdG	=	8-hydroxo-2'-deoxyguanosine
AD	=	Alzheimer's disease
ALL	=	acute lymphoblast leukemia
ALS	=	amyotrophic lateral sclerosis
APOE	=	apolipoprotein E
ATM	=	ataxia telangiectasia mutated protein
ATP	=	adenosine triphosphate
ATR	=	ataxia telangiectasia and Rad3-related protein
BER	=	base excision repair
CNS	=	central nervous system
DAR	=	differentiation-associated repair
FTD	=	frontotemporal dementia
GGR	=	global genome repair
GTP	=	guanosine triphosphate
HD	=	Huntington's disease
IRC	=	individual repair capacity
LDL	=	low-density lipoproteins
LONDD	=	late-onset neurodegenerative disease
MAPT	=	microtubule-associated protein tau
MND	=	motor neuron disease
NER	=	nucleotide excision repair
NMDA	=	N-methyl-d-aspartate
NMDAR	=	NMDA receptor
PCNA	=	proliferating cell nuclear antigen
PD	=	Parkinson's disease
ROS	=	reactive oxygen species
TCR	=	transcription-coupled repair
VCI	=	vascular cognitive impairment.

Pathogenesis of late-onset neurodegenerative diseases: nature vs. nurture

The Gods themselves cannot recall their gifts.

Alfred, Lord Tennyson, “Tithonus” (1859)

Late-onset neurodegenerative disease (LONDD) is characterized by progressive loss of functionality of central nervous system (CNS) neurons and/or neuronal death with age of onset typically after the sixth decade, although the age of onset may vary from 40 to >90. There may be significant variance in their clinical course (from slowly progressive to rapidly progressive), heritability (from monogenic to multifactorial), and penetrance in carriers of known pathogenic mutations (from virtually 100% to below 65%). Most presently available treatments may, at best, alleviate the symptoms of the disease but cannot significantly delay its onset, slow it down significantly, or reverse the pathological changes that have already occurred. Alzheimer's disease (AD) and Parkinson's disease (PD) are the most common LONDDs. These two alone account for a significant proportion of disability after the age of 60 (reviewed in [1,2]). The prevalence of neurodegenerative disease increases with age. In individuals aged 65–75 the prevalence of AD varies between 1% and 5%, whereas in the age range of 75–84 it may increase to 25%–40%.[2,3] Similarly, the prevalence of PD may increase from less than 1% in ‘younger old’ to 5%–10% in those aged over 75.[4] Notably, the prevalence of AD and PD tends to decrease in the ‘oldest old’ (>85 years of age).[2,5]

The genetic basis of LONDD is diverse. Some of the familial forms of amyotrophic lateral sclerosis (ALS1, associated with mutations in the SOD1 gene) and the relatively rare Huntington's disease (HD, due to pathological expansion of a CAG repeat in the HTT gene) are transmitted in an autosomal dominant fashion with virtually complete penetrance, although the age of onset may vary significantly even among patients with the same mutation, suggesting the existence of modulating factors.[6,7] Most LONDDs, including AD and PD, however, are believed to have a multifactorial genesis, as 70%–90% of all cases are sporadic and age-matched asymptomatic carriers of known disease-associated mutations have been described.[8,9] No single environmental factor has been yet implicated as a factor in the pathogenesis of sporadic LONDD with any degree of certainty. Thus, reliable prophylactic measures to delay or prevent the development of LONDD have not been defined yet, apart from the common recommendations given to people over 50 in order to prevent vascular disease, e.g. maintenance of arterial pressure, blood glucose and low-density lipoprotein (LDL) cholesterol levels below threshold values, regular age-appropriate physical activity, use of preventive anticoagulation, etc. This stems mostly from the concept that pre-existing vascular disease and, specifically, stroke are predisposing factors for development of neurodegenerative disease.[10,11] Nevertheless, no data presently available show that preventative measures normally recommended in order to decrease the risk of vascular disease may have any significant effect on the risk of LONDD and the potential outcomes.

Specific neuropathology has been described for the two most common LONDDs: AD (beta-amyloid and tau protein depositions) and PD (intracellular alpha-synuclein inclusions and alpha-synuclein-laden dystrophic neurites). However, the presence of the pathoanatomic hallmarks of AD or PD is not uncommon among the neurologically intact aged individuals, and vice versa – not all clinically recognizable cases of LONDD present at autopsy with the characteristic anatomic findings.[12,13]

A common feature of LONDD is mass neuronal death. It may affect selected neuronal populations (the dopaminergic neurons in substantia nigra in PD; the medium spiny projection neurons of the striatum in HD; the motor neurons in the spinal cord, the brainstem and the brain in motor neuron disease (MND)), or larger parts of the brain (generalized brain atrophy – in AD and frontotemporal dementia (FTD)). At the same time, evidence of attempted re-entry into the cell cycle may often be found in the remaining neurons on the lesion site. Apparently, dysregulation of the mechanisms that govern cell death and cell replacement is part of the clinical picture in LONDD. Below, we briefly review the basic features in the life cycle of adult neurons, including their death and potential replacement in order to set the basis for speculation about the genetic bases of LONDD.

Birth and death (and re-birth) of neuronal cells in LONDD

Cell death is quite normal in the tissues of the multicellular body. Aged or damaged cells or cells that served a purpose during earlier developmental phases but are no longer needed are normally removed from the cell pool. Neuronal tissue in specific brain regions (e.g. the superior colliculi in the midbrain, the visual cortex and others) has physiologically increased rates of cell death during certain phases of development (e.g. in the early postnatal period).[14,15] The rate of cell death in different tissues is dependent on many factors, physiological (preprogrammed pattern of aging for the particular cell type) as well as pathological (toxic effects, trauma, acute or chronic inflammation, etc.).

Many types of differentiated cells in the adult body are normally in a permanent state of replicative quiescence – that is, their programme for proliferation has been blocked at some point prior to completing the differentiation. These are usually highly specialized cells dedicated to a specific function (terminally differentiated cells). Cell division occurs rarely in terminally differentiated cells, although it might be possible in some type of cells and/or under specific circumstances. In most tissues, the cells that have been lost are replaced by new differentiated cells produced by the designated adult stem cells of the tissue. The rates of replacement may vary in different tissues. Some types of differentiated cells such as epidermal cells or erythrocytes are replaced often (within days or weeks). Other types of cells (e.g. the cells of the liver and kidney parenchyma) have significantly slower turnover, in the order of years. The capacity to replace damaged or lost cells generally declines with age, but functional adult stem cells have been isolated from tissues of individuals at any age, including the elderly.[16,17] In the adult CNS, different types of cells may have very different profiles with regard to their turnover rates and the capacity for potential replacement. The glial cells in the adult CNS are replaced regularly. CNS neurons, however, are replaced very infrequently. For several decades, it was believed that no new neurons whatsoever were produced in the adult CNS, so that their number could only decrease with time. Indeed, some neural populations tend to dwindle with age. For example, the total number of neurons in the mediodorsal thalamic nucleus in the newborn is about 40% higher than that in the adult brain.[18] Similarly, a modest decrease of neuronal mass in the hippocampus was noted as a feature of physiological aging in rats and in humans.[19–21] Later, it turned out that replacement of CNS neurons was possible, albeit in a limited fashion. New differentiated neurons are produced by the neural progenitors in the adult neural stem cell niches of the CNS.[22] The neurons of the olfactory bulb, the subventricular zone, the cerebellar cortex and the hippocampus in the adult mammalian CNS are regularly (albeit rarely) replaced throughout adult life.[23–26] In fact, the neurons in the human olfactory bulb are being replaced about every 40 days.[25] There is evidence that neocortical neurons may also be replaced. The total number of cortical neurons in the newborn is virtually equal to the number in healthy adults, although it is known that many neurons in the neocortex die in the course of normal brain development.[27] Also, the neuronal density in the human visual cortex decreases in the first months after birth, remains relatively stable throughout adulthood and actually increases with aging.[15]

Mammalian neural progenitors have inconspicuous morphology with astrocyte-like phenotype but may be distinguished by their expression profile (SOX2(+) IGF2(+) CD133(+) CD24(−) CD34(−) CD45(−).[28–30] Neurogenesis in the CNS may continue well into advanced age, as cells expressing markers characteristic of neural progenitors were found in autopsied human brains of middle-aged and elderly patients with ischemic injury (56−81 years of age) as well as in brains of age-matched controls.[31] Indeed, in the cited study, the absolute number of neural progenitors was reported to be lower in older brains than in younger brains, but still, renewal of cells was apparently possible in advanced age. Ongoing physiological neurogenesis was demonstrated in the substantia nigra pars compacta of neurologically intact adult mice.[32] In the cited study, the rate of generation of new neurons in substantia nigra was found to be significantly lower than the rate of neurogenesis in other zones where neurogenesis was normally observed (specifically, the hippocampus) but still, the observed rate of neurogenesis in substantia nigra was estimated to be potentially sufficient to replace all of the dopaminergic neurons within the average mouse lifespan.[32] Generally, the production of differentiated cells by the adult stem cell niches can be upregulated, e.g. in the presence of tissue damage (trauma, inflammation, etc.). This is apparently valid for neuronal tissue as well, as enhanced neurogenesis was demonstrated in brains of mice with parkinsonism induced by MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, a compound known to produce severe and lasting Parkinson-like symptoms due to damage to dopaminergic neurons in the midbrain).[33] Similarly, induced loss of motor neurons in the spinal cord of mice modelling MND was shown to stimulate the proliferation of resident neural progenitors.[34] In the acute settings, transient upregulation of the progenitor cells of the tissue (e.g. in the presence of serious tissue damage) is beneficial for the organism, as long as it is downregulated back to its baseline level after the integrity of the tissue is restored. In the long term, however, chronic upregulation of the proliferation activity of the stem cell niche may have harmful effects. The latter may vary in different tissues, depending mainly on the proliferation capacity of the resident cell niche. Generally, adult progenitor cells are capable of division, but the number of divisions is finite. The proliferation capacity of adult progenitor cells is limited by the mechanism of physiological (preprogrammed) cell aging, a mechanism ensuring that cells that have sustained too much damage would not divide further. In tissues where the resident adult stem cell population has significant proliferative potential (e.g. epithelium), overstimulation of the proliferative capacity of the stem cell niche and the subsequent rapid cycling may result in gradual accumulation of mutations with every round of division. The latter may significantly accelerate the normal process of tissue aging and/or trigger abnormal cell growth (carcinogenesis). Indeed, if the cells newly produced by the adult stem cell niche are capable of multiple divisions in the course of their differentiation (as are, for example, the precursors of blood cells), the increased demand for new differentiated cells may not contribute significantly to the acceleration of the onset of replicative senescence in the adult stem cell niche, as the increased need for differentiated cells may be managed by stimulation of the division of the partially differentiated precursors. However, in cases where the differentiation programme includes blockade of the proliferation programme early in the course of differentiation (as is the case with neural cells), the increased demand for new differentiated cells may only be covered by ongoing stimulation of the stem cell niche to produce new differentiated cells. In tissues where the proliferative potential of adult stem cell progenitors is a priori limited (neuronal cells, cardiac muscle cells), overstimulation to produce new differentiated cells is more likely to accelerate the onset of replicative senescence than trigger tumorigenesis (indeed, primary neuronal tumours such as neurocytoma and gangliocytoma and cardiac rhabdomyomas are quite rare and generally unaggressive). Increased rates of neurogenesis have been observed in brains affected by neurodegenerative disease. Increased hippocampal neurogenesis was observed in transgenic AD mouse models was determined to be a major hallmark of the progressive stage of AD in the mouse.[35]

Replenishment of the stores of differentiated cells of the tissue by overstimulation of the adult stem cell niche may maintain the integrity of the tissue for a long time. Eventually, the neural stem cell niche may become incapable of producing new differentiated cells to match the neuronal loss. Multiple studies demonstrate that impaired neurogenesis in the CNS (reduction in the proliferation capacity of neural progenitor cells, dysfunctional neuronal differentiation and abnormally upregulated neurogenesis) may be among the early events in the pathogenesis of AD.[36,37] Ectopic expression of marker proteins typical of immature neurons, such as PSA-NCAM, NeuroD1, NeuroN, TUC4 and doublecortin, indicating for dysfunctional differentiation of neural precursors and/or attempted reactivation of the division programme of adult neurons (see later) was observed in rat models and in brains of patients with AD.[37–39] The areas where neurogenesis continues into adult life are often among the areas where neuronal loss becomes apparent early in the course of LONDD. The hippocampus is among the areas that experience mass neuronal loss very early in AD.[20,21] Reduced capacity for neurogenesis in the subventricular zone was observed in transgenic AD mice.[36] It is also a well-known clinical fact that olfaction is affected early in the course of many LONDDs (PD, AD, FTD).[40–42] The olfactory bulb is believed to be among the first brain regions affected by alpha-synuclein deposition in PD and decrease in the olfactory bulb volume is a common finding in autopsied brains of patients with PD.[43,44] It is possible that the phenomenon of increased neurogenesis seen in LONDD may be an attempt to compensate for the increased rate of neuronal loss. The latter, however, may only provide temporary respite, and is likely to fail, eventually, albeit after many years (see later).

Under physiological conditions, differentiated CNS neurons are incapable of further division. Re-entry of differentiated neurons into the cell cycle is described as a characteristic part of the pathology in LONDD. Evidence of attempted reactivation of the division programme (increased amount of DNA across multiple loci indicative of replication, dysfunctional segregation of chromosomes, aneuploidy) has been repeatedly observed in autopsied brains of mice modelling familial AD and of human patients with AD, PD and ALS.[45–49] The levels of some of the key molecules responsible for cell cycle progression (Cdk4 and its pairing partner cyclin D, Cdk1/cyclin B, phosphorylated retinoblastoma protein and E2F) – in other words, markers of abnormal cell cycle re-entry – were found to be significantly higher in brains of patients with AD than in age-matched controls.[50]

Want to live longer? Change your attitude. Mechanisms employed by differentiated cells to increase or decrease the lifespan of specific cell populations

One may expect that physiological (age-dependent) or pathologically accelerated replicative senescence of the adult stem cell niches would, eventually, result in degenerative disease in most types of adult tissues. Whether this would occur sooner or later would depend on the rate of depletion of the niches. Neuronal tissue has one specific feature that may, under physiological conditions, significantly delay cell loss and, respectively, the need for cell replacement. This feature is their exceptional longevity, comparable only to the longevity of adult cardiac muscle cells (the latter, however, have a better pronounced regenerative capacity). Terminally differentiated neuronal cells in the adult body are naturally programmed for a significantly increased lifespan compared to other cell types in the multicellular body. Thus, there is a high likelihood that adult neurons would function throughout the life of the organism without need to be replaced, or that the need for replacement would occur very rarely.

The exceptional longevity of adult neurons is ensured by two linked mechanisms. The first mechanism (clockdown mechanism) slows down the unidirectional clock measuring the lifespan of a progenitor cell with the number of cell divisions. This is the universal mechanism ensuring that the progenitor cells in the adult stem cell niche need to divide only infrequently.[17] We propose that there is a second mechanism (threshold modulation mechanism) ensuring that the threshold beyond which damaged cells are routed to apoptosis is significantly higher than the threshold in other types of cells (for details, see later). The threshold modulation mechanism comprises two essential components:

1. Modulation of the capacity to identify/repair cell damage (in neurons, mainly oxidative damage) so that oxidative damage is promptly repaired as long as it is in regions that are important for the functioning of the neuron (specifically, transcribed regions).

2. Modulation of the capacity for assessment of the scope and scale of genomic damage so that not all data is included in the assessment of damage.

In physiological settings, these two mechanisms work together to ensure that the majority of the adult CNS neurons would live until the organism lives, setting aside a small potential reserve of differentiated cells. The clockdown mechanism is based on slowing down physiological aging of the adult stem cell niche (already discussed above). It is a common mechanism for all adult progenitors, although the time interval between two consecutive divisions of the progenitor cell may strongly vary between different cell types. The threshold modulation mechanism may exist in two subtypes and is generally used to delay/avoid or to accelerate the death of specific cell populations, depending on the needs of the organism. It is employed only by several types of cells in the adult body – specifically, terminally differentiated neurons, but also other types of cells. The results may be different depending on the particular subtype of the mechanism and the type of the cells. Notably, both versions of the threshold modulation mechanism are employed only by cells that are unlikely to divide further.

To live is to die. Evidence of the role of oxidative damage in LONDD

About 20%–25% of the energy produced by the cells of the human body is consumed solely by the brain (which makes up only 2%–2.5% of the body mass). Neurons require high amounts of adenosine triphosphate (ATP) for the purposes of maintenance of membrane potential, depolarization and repolarization, synaptic and extrasynaptic signalling. Oxidative phosphorylation is inevitably associated with production of large amounts of reactive oxygen species (ROS): superoxide radicals, free hydroxyl radicals, singlet oxygen, etc. Oxidative damage (affecting cellular proteins, lipids and DNA) is the major type of damage in neural cells. Evidence of increased oxidative damage has been repeatedly observed in specific neuronal populations and/or in larger areas of autopsied brains of patients with LONDD. Markers signifying increased levels of oxidative stress have also been repeatedly observed in vivo in the brains of double-mutant (expressing mutant human APP and PSEN1) and triple-mutant (APP, MAPT and PSEN1) transgenic mouse models of AD.[51,52] Different studies have identified 3- to10-fold difference in the levels of oxidized bases in autopsied brains of patients with AD than in age-matched controls.[53,54] Persisting oxidative damage apparently plays a role in the pathogenesis of AD-like dementia commonly developing in patients with Down syndrome that have survived childhood and adolescence. In fibroblasts from adult patients with Down syndrome, the amount of oxidized bases (specifically, dihydro-8-oxoguanine (8-oxodG)) was increased compared to healthy age-matched controls and the levels of base excision repair (BER) proteins such as XRCC1 and DNA polymerase beta were upregulated.[55] Increased levels of 8-OHdG (8-hydroxo-2'-deoxyguanosine) and H2AX (a variant of the core histone H2A, one of the earliest damage-signalling molecules) were found in brains of elderly (85 years of age) patients with AD as well as in age-matched controls without cognitive decline, but the levels of damage in the AD group were significantly higher than those in the control group.[50] In mice deficient for MTH1 (8-oxo-7,8-dihydroguanosine triphosphatase, a deoxyguanosine triphosphatase (deoxy-GTP-ase) that hydrolyses oxidized guanosine triphosphate to monophosphate, thus minimizing the risk of incorporation of 8-oxoguanine in DNA), treatment with MPTP produced neurological dysfunction that was significantly more severe than in wild-type mice.[56,57] Later, it was shown that mice null for the Ogg1 gene (coding for a specialized BER glycosylase for removal of 8-oxoguanine) exhibited age-associated loss of the nigrostriatal neurons and increased sensitivity to MPTP.[58] Even though HD is linked to a specific gene (HTT in humans, Hdh in mice), Ogg1 was also shown to be directly involved in age-dependent somatic expansion of triplets in the Hdh gene in mouse models of HD.[59] Studies of the role of oxidative damage in PD conducted in human patients confirmed the findings from the murine models. Elevated levels of 8-OHdG (an oxidized derivative of deoxyguanine) were found in the mitochondrial DNA of cells from the substantia nigra pars compacta of brains of patients with PD.[60,61] Later it was found that the expression of hOGG1 and MUTYH (a key enzyme from the mismatch repair system) was upregulated in brains of patients with PD.[62,63] It is likely that this upregulation occurred as an attempt to manage increased genotoxic stress. In foetal and adult brains of patients with Down syndrome, the level of p53 was significantly increased and the major BER factors, such as XRCC1 and DNA polymerase beta, were predominantly bound to chromatin even in the absence of DNA damage.[55] This was interpreted by Necchi et al. [55] as a mark of abnormal persistent activation of the DNA damage response pathways/BER deficiency as it was accompanied by phosphorylation of H2AX (associated with relaxation of the chromatin structure in the vicinity of the damage so that the cell machinery may have unimpeded access) and checkpoint protein kinase Chk2 (stabilizing p53 by phosphorylation).[64,65] Increased levels of DNA damage (specifically, oxidative damage) have been found not only in neurons of patients with AD and PD, but also in tissues completely unrelated to the CNS such as peripheral leukocytes.[66,67]

Fix it or die. Potential outcomes of assessment of the integrity of the genome in damaged cells

In most mammals (with the notable exception of rodents), the cell machinery for identification and repair of DNA damage would actively seek out and repair damage in all regions of the genome. Indeed, priority is normally being given to transcribed regions and the transcribed strands of expressed genes.[68,69] Still, virtually all damages would register and, eventually, all damages would be repaired, albeit in its own time. If the cell could not handle the damage for some reason (damage is too great and/or too extensive and/or the repair machinery is not working at nominal capacity), some of the damages may remain unrepaired. In this case, the integrity of the genome would be assessed again in order to prevent damaged cells from further division. Generally, the outcomes of this ‘post-repair’ assessment may be as follows:

1. Most of the damage has been identified and repaired; levels of residual unrepaired damage are low. The cell may be assessed as ‘good to go’ and allowed to proceed further with cell division, if its programme includes division.

2. There is still considerable amount of unrepaired damage. Another round/s of damage identification and repair are carried out and further cell division is postponed until assessments show that the level of residual damage has fallen below a certain threshold. This may occur at some point (if the damage is eventually repaired), or may never occur. In the case of the latter, the cell remains permanently locked in a state when division is no longer possible (replication senescence) and may, after some time, die.

3. If the damage is assessed as too great and/or too extensive to manage, attempts at repair may not be undertaken and the cell may be routed directly towards the programmed cell death pathway.

The outcome of the assessment of genomic integrity (‘pro-repair’ or ‘pro-apoptotic’) depends on many factors, including the intensity/duration of damage and the capacity of the cell machinery for identification and repair of damage to handle the damage. The efficiency of the repair mechanisms generally declines with age, thus genotoxic damage that could easily be handled in young age may cause considerable difficulties in advanced age. Transformed cells may easily tolerate much higher amounts of damage than normal cells provided that the transformation included abrogation of major mechanisms for imposing cell cycle arrest/programmed cell death in the presence of damage. Differentiated neurons may also tolerate higher amounts of damage than other types of cells and may delay or evade programmed cell death. Their strategy also involves tampering with the systems for assessment of damage, although in a manner that is dissimilar to the strategy used by cancer cells.

The ultimate outcome engineers. Differentiated cells may manipulate the assessment of damage in order to modify the decision about the fate of the cell

The assessment of damage has its specificities in dividing and non-dividing cells. In dividing cells the assessment of damage is specifically rigorous in the presynthetic phase (G1) of the cell cycle, although checks are also carried out in the S, G2 and M phases of the cell cycle.[70] Every subsequent division increases the risk that upon the next assessment the cell would have sustained unrepaired damage and would have to be decommissioned by replicative senescence and/or cell death. In cells that do not normally divide (e.g. terminally differentiated cells) assessment of genomic damage is still regularly carried out (although in different manner in different types of cells) and damaged cells are still rerouted to apoptosis if the scale and scope of damage is assessed as being too great to manage. In general, higher levels of genotoxic stress (or lower capacity of the cell to handle genotoxic damage) would potentially accelerate cell decommissioning due to accumulation of damage beyond a certain threshold. Some types of differentiated cells, however, may be specifically adept at manipulating the amount of damage included in the overall assessment so that it is either underestimated or overestimated in order to delay or accelerate cell death. Terminally differentiated neurons routinely manipulate the input in the damage assessment in order to alter the prospective decision about the fate of the cell.

It was already mentioned that damage in transcribed parts of the genome is repaired with priority. Still, most mammalian cells would eventually check and repair all damages in their DNA. Thus, damage in untranscribed regions counts in the assessment of the integrity of the genome, even though the cell machinery only gets to repair it after it has finished with damage in transcribed regions. The location of damage within the genome is apparently so important that one of the mechanisms for DNA repair (nucleotide excision repair (NER), the most versatile and the most widely employed among all DNA repair mechanisms) has developed two distinct subtypes dealing with damage in transcribed regions (transcription-coupled repair, TCR) and in nontranscribed regions of the genome (global genome repair, GGR). These two subtypes employ essentially the same mechanism of repair of DNA damage (nucleotide excision and strand resynthesis) but have different fields of action (the transcribed and the untranscribed DNA); are triggered by different signals (TCR, by RNA polymerase II stalled at the damage site; GGR, by localized strand distortion caused by DNA damage); and may be differentially regulated (selectively up- or downregulated). In mature neurons, GGR is selectively suppressed and virtually no repair is carried out in the untranscribed regions of neuronal DNA.[71,72] Effectively, all repairs carried out by nucleotide excision are concentrated on the transcribed regions. Thus, all of the considerable capacities of the NER machinery in differentiated neurons are dedicated solely to the maintenance of the integrity of a relatively small part of the genome – specifically, the regions containing the genes coding for proteins needed for the survival and the functioning of the neuron.

In differentiated neurons, there is also another level of prioritization of repair in transcribed over untranscribed regions. Specifically, the untranscribed strand in transcribed regions is also checked and repaired with priority (differentiation-associated repair, DAR), whereas in other types of cells it is generally repaired together with untranscribed portions of the genome.[72,73] This is another mechanism to ensure the integrity of the transcribed parts of the genome at the expense of the untranscribed DNA. Due to the synchronized functioning of those two mechanisms and the fact that the risk of replication-dependent introduction of errors in DNA is virtually non-existent in neurons, the risk of occurrence of unrepaired damage in the transcribed genes of a normal neuronal cell would be very low (although it may increase as the organism ages due to the inevitable decline in reparative capacity). Damage in the untranscribed regions of the genome of differentiated neurons that is reparable by specific mechanisms other than NER (oxidised bases, base mismatches, strand breaks, etc.) would eventually be repaired, after the repair in the transcribed regions is complete. Nevertheless, the mechanisms of NER would not take part in the identification and repair of damage in untranscribed DNA, although NER may repair almost all types of damage. Thus, in untranscribed regions, the checks for the presence of damage would only be carried out by the designated mechanisms of BER, mismatch repair and recombination. Every repair mechanism is essentially ‘leaky’, that is, it may occasionally miss damage or may introduce errors in DNA due to the need of DNA synthesis for completion of any type of repair. The targets of different mechanisms for identification and repair of errors in DNA generally overlap, probably in order to limit the error rate in the course of DNA repair. In cells in which the most versatile system for identification of damage in DNA – that is, NER – has been disabled for the larger part of the genome, the risk that residual damage may remain undetected and unrepaired in the untranscribed parts of the genome is generally higher than in most cells. It also means, however, that the overall amount of damage included in the assessment of genomic integrity may be lower, as the data input to the functioning detection systems would not be complemented by the input of NER. Thus, it is likely that, in differentiated neurons, the amount of damage that registers in the assessment of genomic integrity is lower (probably, significantly lower) than in other types of cells. Of course, this does not decrease the absolute rate at which damage accumulates, but, rather, it reduces the relative amount of damage that counts in the assessment. Thus, under physiological conditions the neuronal cell must sustain significantly higher absolute amount of damage than other types of cells so as to be assessed as irreparably damaged (that is, the threshold beyond which a damaged neuronal cell is targeted towards the apoptotic pathway is higher than that in other types of cells). This threshold may not be reached for a long time (not until late in life, after the age-dependent decline of the repair capacity has become significant) or may not be reached at all within the lifespan of an individual. However, if there are factors that increase the genotoxic impact onto the neural cell, the threshold may be reached sooner. Slowly accumulating DNA damage that eventually triggers neuronal cell death could explain the age-dependent trend in the prevalence of sporadic PD and AD (steadily growing up to the age of 85). Evidently, the longer a factor acts and the weaker the mechanisms that identify and repair the damage become, the closer the cells are to the threshold beyond which the cell is routed to apoptosis and, respectively, the higher the risk of development of neurodegenerative disease. It may also explain the fact that the risk of AD and idiopathic PD actually decreases in the ‘oldest old’.

A somewhat similar mechanism for modification of the outcome of assessment of genomic integrity has been shown for other types of non-dividing mammalian cells – namely memory B-cells [74] and monocytes.[75,76] Unlike neurons, in which the outcome modifications serve to extend their lifespan and avoid cell death, the mechanism used by blood cells essentially works the other way round, increasing their vulnerability to cell death. In monocytes, the efficiency of global genomic repair is actively downregulated but the mechanisms that assess genomic damage are upregulated so that cells that have sustained damage would be rapidly redirected towards apoptosis. In this case, the cells employing the threshold modulation mechanism are not terminally differentiated, but their further division is unlikely, as their proliferation programme has already been blocked (monocytes) or the chances that the cell might have to divide ever again are quite low (memory B-cells). In monocytes, the threshold modulation mechanism is employed as a form of control of the locally produced amount of ROS at sites of infection or inflammation. Monocytes, as precursors of ROS-producing macrophages are made specifically vulnerable to damage (including damage by ROS) by downregulation of the expression of key proteins of DNA repair, so that they may be rapidly routed to apoptosis after the infection or the inflammation has been cleared.[76] Failure of this mechanism is associated with increased risk of several human diseases with multifactorial genesis (rheumatoid arthritis, inflammatory bowel disease and others).[77] Similarly, in B-memory cells that have their global repair capacity inactivated, activation towards proliferation is accompanied by accumulation of damage so that the cells may be routed towards cell death after the factor that has caused the activation of the memory B-cell clone has been eliminated. Increased rates of somatic mutagenesis in activated B-cell clones have been implicated in the pathogenesis of B-cell lymphoma.[74]

Apparently, the outcome of the assessment of the integrity of the genome for some types of non-dividing cells may be significantly different than the outcome for other cell types. In neuronal cells, this is a mechanism extending the life of the cell and, at the same time, preserving its functional capacity. It might be possible that the molecular origins of neurodegeneration are associated with dysregulation of the mechanism for modification of the outcomes of the assessment of the integrity of the genome in neural cells.

A theme with variations. Effects of differential regulation of components of individual repair capacity in neuronal tissue

There are two essential components of the nature-made mechanism that protects eukaryotic genomes from accumulating too much damage: a mechanism for identification and repair of genomic damage and a mechanism for assessment of genomic integrity. The latter includes the integrated sub-mechanism that sets the threshold beyond which the cell fate is dramatically altered (induction of cell cycle arrest and/or rerouting to apoptosis). We already showed that this threshold may be actively manipulated in neurons and other non-dividing cells in order to modify the outcome of the assessment of genomic integrity. However, this does not change the absolute amount of damage that the cell sustains every day. Generally, all cells in multicellular organisms have to manage a very large amount of damage (in the order of 10⁴ events/day).[78] Considering that the average mutation rate in mammalian cells is only about 10⁻⁹ per base per generation, the machinery for identification and repair of damage apparently works fairly efficiently, at least in young age. Nevertheless, its efficiency (in terms of absolute amount of damage repaired in a fixed interval of time or the amount of residual damage that had remained unrepaired after a fixed amount of time) may vary significantly even between age-matched clinically healthy individuals. The capacity to handle DNA damage and maintain genomic integrity is commonly referred to as ‘individual repair capacity’ (IRC), although the term may underestimate the importance of the maintenance of the genomic integrity. The two basic components of IRC function together and may significantly influence each other. Intact functioning of the one component may (at least temporarily) compensate for potential deficiency of the other component. Indeed, a lower-than-normal capacity to identify and repair genotoxic damage may be temporarily compensated by increased propensity to destroy cells that have sustained too much damage (a ‘pro-apoptotic’ tendency), although this may cause trouble in the long run. Similarly, a very efficient mechanism for DNA repair may result in very low level of residual DNA damage, so that even in cases when the cell is prone to repair damage rather than kill damaged cells (a ‘pro-repair’ tendency), errors in DNA would accumulate very slowly (again, this may work very well in young age and may be a source of significant issues in old age). Only in case both of those two principal components of this mechanism fail (e.g. in advanced age, or under conditions of increased genotoxic stress), abnormal cell growth or abnormal death may occur. It is difficult to argue which of the two potential pathways after assessment of genomic integrity (‘pro-repair’ or ‘pro-apoptotic’) is better or safer. On the one hand, ‘pro-repair’ tendencies may increase the risk of propagation of carcinogenic mutations. On the other hand, a propensity to direct damaged cells to a pro-apoptotic pathway rather than attempt to repair may accelerate cell and tissue aging and trigger degenerative disease. It might appear safer if all damaged cells were routed towards apoptosis, as it would decrease the risk of carcinogenic transformation. This, however, may work only in tissues where the resident adult progenitor cells are capable of promptly replacing lost cells with a rate to match the rate of loss (the latter, however, may also become less efficient as age advances). One could hypothesize that differentiated neural tissue presents a case in which the propensity to retain damaged cells may be significantly more favourable for the long-term outcome than the propensity to remove damaged cells from the cellular pool. Of course, keeping and repairing cells with damaged DNA has the inherent risk of accumulation of mutations and their potential transmission to the cell's progeny. In neurons, this is compensated by disabling the division programme and concentrating the capacity of NER only on transcribed regions. The latter has two potential advantages: (1) it reinforces repair of regions that are important for the functioning of the cell, ensuring that the residual level of damage would be very low; and (2) not all damages are factored out in the assessment of genomic integrity so that it is unlikely to reach the threshold beyond which the cell is directed to apoptosis. Thus, despite the high levels of damage that neurons sustain throughout their very long lives, they may escape programmed cell death for decades – probably until accumulated damage interferes seriously with their specialized functioning. The latter is more likely to occur in advanced age, when the efficiency of repair (including the reinforced repair of transcribed regions) declines.

We hypothesize that the physiological mechanism for maintenance of neuronal longevity may be based on a combination of the following two components:

1. improved detection and/or repair of damage in neural cells, limiting the damage that neurons sustain in the course of their functioning; and/or

2. relaxed or otherwise modified mechanism of assessment of genomic integrity, so that cells that have sustained damage are retained (at least temporarily) and subjected to repair of damage rather than targeted towards the programmed cell death pathway.

According to this model, the underlying molecular pathology in common sporadic LONDD may involve a combination of genetic traits conferring subtle but long-standing deficiency in repair of genotoxic damage and a propensity to assess damage as irreparable and to target damaged cells to apoptosis. As a matter of fact, this combination may hypothetically explain late-onset degenerative disease of any type, but neural tissue would be specifically vulnerable because of its limited capacity to replace lost cells. The two factors ought to work together for a long time (potentially–since birth) in order to significantly increase the risk of development of neurodegenerative disease several decades later.

Let us consider the situation in which the neural tissue is subjected to long-term genotoxic stress that it could barely handle because of a lifelong subtle deficiency in mechanisms of detection and repair of damage in DNA. One could expect that the absolute amount of damage would increase at a faster rate than in cells with average or near-average capacity to repair damage. Accumulation of mutations over a certain threshold would, eventually, result in triggering of the programmed cell death cascade. If the cells are genetically predisposed towards routing damaged cells towards the apoptotic pathway, mass neuronal death may begin earlier in life. Increased rates of neuronal cell death may lead to continuous activation of the stem cell niche in order to replace the cells that have been lost. However, the proliferative capacity of adult neural stem cell niches is limited. Thus, the moment when the progenitor cell that has been forced into continuous division enters replicative senescence may come sooner than what is normal for the cell type. This may result in premature depletion of the stem cell niche. Here, the term ‘depletion of the niche’ does not mean that the number of the progenitor cells in the niche would decrease significantly (although some of them may die) but, rather, that their capacity to divide further would decrease with every subsequent division. As more and more divisions occur, the potential of the progenitor cells to produce differentiated cells would eventually decline, so that the regenerative capacity of the tissue progenitors may not be able to keep pace with the rate of cell loss and the rate of replacement would slow down. This may be the beginning of degenerative disease. As more time passes, the rate of cell death would not slow down, while the rate of regenerative capacity of the tissue progenitors would decline even further. Eventually, they would become irresponsive to signals for further stimulation. In the end, a point may be reached where the integrity of the tissue would depend directly on the lifespan and the functioning of the latest ‘batch’ of differentiated cells. After the latter have died, there would be virtually no functional cells left and the cell niches would be incapable of producing more, which would be the end state (overt degenerative disease). Theoretically, after the proliferative potential of the adult stem cell niche is depleted, new differentiated cells could be supplied by reactivation of the division programme in differentiated neurons. Very few specialized cell types, however (hepatocytes, smooth muscle cells, skin fibroblasts), are capable of physiological exit from replicative quiescence and re-entry into division. Attempted reactivation of the cell cycle in differentiated neurons usually results in gross abnormalities that eventually trigger the cellular mechanism of apoptosis of damaged cells. In the end, the neurons that attempted to divide in order to compensate for accelerated neuronal loss would also die.

It is possible that either of the essential factors (decreased repair capacity/increased propensity for routing damaged cells towards apoptosis) may eventually produce mass neuronal cell death on their own. It could be expected that the threshold beyond which the cell fate is dramatically altered may be reached sooner under conditions of unmanageable genotoxic stress. It may be related to environmental causes (exposure to potent genotoxic agents, acute incidents producing increased levels of ROS, e.g. vascular incidents) or to endogenous causes (increased flow of ROS produced by mitochondria, e.g. associated with severe lasting hyperglycaemia, hyperlipidaemia, structural alterations in the mitochondrial DNA, etc.). After all, neurotoxicity is a common complication of uncontrolled diabetes and some authors believe that diabetes may be a predisposing factor to development of all types of dementia, including AD (supported by the finding that glucose-lowering medication and insulin may ameliorate the phenotype in murine models of AD.[79–81] Carriership of APOE4 alleles (associated with increased levels of plasma lipids) is a major genetic factor increasing the risk of late-onset AD, although the functions of apolipoprotein E (APOE) are not directly related to amyloid protein precursor and tau protein processing.[82] In order to produce the phenotype of degenerative disease, however, the duration and/or the intensity of the genotoxic impact ought to be significantly higher than the average daily amount of damage; or the cell machinery for detection of repair of damage ought to be partially or completely disabled. Human-inherited disease due to severe molecular defects causing deficiency of DNA repair is almost always associated with severe neurological impairment (an exception is xeroderma pigmentosum complementation group C (XP-C) where only global NER is affected). In this case, however, the deficiency in DNA repair is profound, affecting the functioning of many other tissues besides differentiated neurons in the CNS. In other words, daily genotoxic stress in cells that are only subtly deficient in DNA damage identification and repair may not trigger the vicious circle of neuronal death/overstimulation of the adult progenitor niches/depletion of the niche on its own, unless it is complemented by increased propensity for apoptosis of damaged cells (at least, not until very old age, when other issues (e.g. vascular incidents) may accelerate neurological decline).

Increased propensity for routing damaged cells towards apoptosis on its own may also be incapable of triggering late-onset neurodegeneration. Inherently high levels of pro-apoptotic molecules may accelerate cell death in selected cell populations, including differentiated neurons. A dramatic elevation of the expression of several pro-apoptotic proteins (Bax, Fas, p53 and others) coupled with downregulation of universal anti-apoptotic proteins such as Bcl-2 and Hsp-70 as well as ‘specialized’ anti-apoptotic proteins, such as the neuronal apoptosis inhibitory protein, was reported in foetal brains with Down syndrome.[83,84] A recent study in Drosophila showed that the activity of the major executor caspase-3 increased in an age-dependent manner in specific neurons involved in olfaction and visual memory, eventually causing neuronal death.[85] However, since neurons manipulate the data input in the assessment of genomic integrity, in the presence of near-normal capacity for damage repair, increased pro-apoptotic tendency may be overruled by the essentially high threshold beyond which the cell is routed to apoptosis. In those with very efficient DNA repair, the threshold of damage beyond which differentiated neurons are routed to apoptosis may be reached very late in life (or not reached at all). In such individuals, the effects of a pronounced ‘pro-apoptotic’ tendency may become evident only after the efficiency of DNA repair has declined significantly – that is, in advanced age, and is more likely to show in other tissues before it affects differentiated neurons. Such individuals may be resistant to neurodegenerative disease in their old age but prone to accelerated aging/degenerative disease in other organs and tissues (predominantly tissues with rapid natural turnover in which the contribution of replication-associated errors may be significant – skin, haematopoietic tissue, etc.). It is likely that only the unlucky combination of decreased capacity for DNA damage identification and repair/increased propensity for apoptosis of damaged cells may trigger premature neuronal death in the middle-aged and ‘younger old’, that is, those in the fifth or sixth decade of life. Each of the two factors alone may be expected to exhibit its effect in an age-dependent manner, with the prevalence of the associated pathology increasing with age, reaching a peak after 70–75 years of age. The two factors together, however, may significantly decrease the age of onset of LONDD.

The genotype–phenotype correlations of carriership of allelic variants modulating individual repair capacity are currently under intensive study. Multiple allelic variants in gene coding for proteins of damage identification and repair and maintenance of genomic integrity have already been described. The individual contribution of each of them may be minute, but taken together, they may constitute a favourable background for development of disease and/or a starting point in establishment of individualized therapy to delay or prevent disease. It is likely that in the near future panels of individual repair capacity may be used to determine the risk of development of common late-onset diseases (cancer and degenerative disease).

Why me? Vulnerability of selected cell populations in LONDD

The search for an answer to the question why selected neural populations in the CNS were more vulnerable to neurodegenerative disease than others began in the late 1990s with the advent of the excitotoxic hypothesis. Its essence was that selective neural cell death may be caused by toxic overstimulation of excitatory amino acid receptors – specifically, glutamate and N-methyl-d-aspartate (NMDA) receptors in the target cells (reviewed in [86]). Numerous NMDA receptor (NMDAR) antagonists were subsequently developed and entered into trials for neurodegenerative disease, stroke and acute brain trauma, failing every single one.[87,88] The basic premise, however – that there was a specific molecular cause for the vulnerability of certain neuronal populations – was apparently correct, although the exact mechanism remained elusive. It is possible that the excitotoxic mechanism may be valid for some neuronal populations and not for others, or that it may be valid for some developmental stages only. Most adult CNS neurons contain NMDA receptors, synaptic as well as extrasynaptic. The majority of the NMDA receptors in developing neurons are extrasynaptic. In vitro studies in differentiating neurons show that about 75%–90% of the NMDA receptors are extrasynaptic after 7 days of culturing and then decrease to 20%–50% after 14 days of culturing.[89] In vivo these percentages may be different, but studies in acutely dissected rat hippocampi revealed 36% extrasynaptic NMDA receptors in neurons in the CA1 subdivision of the hippocampus.[90] It is possible that toxic overactivation of NMDA receptors might play a role in the death of differentiating neurons produced by the adult stem cell niches in the hippocampus (and, possibly, in other locations). The resulting shortage of differentiated cells stimulates the stem cell niche to produce new neurons (which are also likely to die in the process of their differentiation), thereby closing the vicious circle that would eventually lead to depletion of the adult progenitor niche.

The generalized brain atrophy commonly seen in AD may be explained from the viewpoint of the hypothesis that a combination of a subtly defective mechanism for detection and repair of oxidative damage in DNA and increased propensity to route damaged cells towards apoptosis may cause mass age-dependent neuronal cell death and subsequent depletion of adult progenitor niches. Indeed, a genetic factor is likely to affect all cells of the body (or, as is in this case, all cells that employ similar mechanisms for assessment of damage, especially if it involves constitutively expressed proteins). Explaining the mass death of specific cell populations in PD and HD may prove much more challenging. One may suppose that involvement of specific neuronal populations but not others may be dependent not only on the overall capacity for repair of DNA damage/assessment of genomic integrity but also on the different effects conferred by carriership of different allelic variants of the components of the system for repair of DNA damage and assessment of genomic integrity within the genotype of the carrier. Carriership of polymorphic variants in different genes may have differential impact on the individual capacity to handle damage and to decide the fate of damaged cells. Specific combinations of different alleles within the same genotype coupled with environmental factors may, in the course of individual life, increase the propensity of some cell populations to die in the presence of damage whereas other populations may still be able to manage (if only just). It is also possible that the genetic background established by the presence of variant alleles in the genotype may cause impairment of the physiological regulation of the mechanisms of expression of repair proteins, eventually resulting in isolated up- or downregulation of specific repair mechanisms in different cell populations. In 2009, an intriguing correlation between the amount of key BER proteins and the somatic instability of CAG repeats was demonstrated in brains of mouse models of HD.[91] Specifically, it was shown that the activity of DNA polymerase beta (the main polymerase synthesising the missing nucleotide in short-patch BER that may also substitute for polymerases delta and epsilon in long-patch BER) was higher in the striate nucleus (i.e. the affected brain region) than in the cerebellum of HD mice. At the same time, the activity of 5'-flap endonuclease (5'-FEN, the enzyme catalysing the processing of the intermediate structure generated by DNA polymerase-dependent displacement of nucleotides in the 3'-vicinity of the repair site in long-patch BER) and of HMGB1 (a major transcription and DNA repair modulator protein) was much lower in the striatum of HD mice than in other parts of the brain, e.g. the cerebellum. The dysregulation of the activity of DNA polymerase beta and 5'-FEN was observed only in the striatum of HD mice but not in other parts of the brain. The authors hypothesized that an overactive DNA polymerase beta created excessive amounts of 5'-flaps at CAG repeats that were not removed effectively because of the underactive 5'-FEN, thereby promoting further (somatic) instability of the CAG repeat. The cause for the differential regulation of the expression of these two repair proteins only in select parts of the mouse brain was not identified, but the results of this study supported the concept of differential modulation of the repair capacity in different cell populations. Recently, it was shown that carriership of a mutation in the murine DNA polymerase beta gene conferring lower copying fidelity and significantly decreased (∼20-fold) steady-state catalytic rate of the enzyme was associated with reduction of the triplet expansion frequency on the somatic level as well as on the germline level in fragile X mouse models.[92] Apparently, the separate factors constituting the individual repair capacity may have different impact in different cell types and in different subpopulations of the same cell type, the effect being probably modified by other genetic factors and factors from the environment. One could expect that the effects of dysregulation of DNA damage identification and repair would increase under conditions of excessive genotoxic stress or in advanced age, when the capacity for repair of damage begins to decline. Some mechanisms of DNA damage detection and repair may fail more rapidly than others, as every genotype contains a mixture of genetic variants conferring subtly lower, nominal or, in rare cases, a slightly increased activity of the encoded protein. It may only be through interaction of the genotype (decreased capacity to handle damage/increased capacity to route cells towards apoptosis/increased endogenous influx of ROS) and the environment (acute or chronic exposure to genotoxic agents) that the eventual outcome for the particular tissue and the particular cell population is determined.

Other factors, including mutant or otherwise altered mRNA and proteins may modulate the effects of pro-apoptotic factors, decreasing the threshold for neuronal apoptosis and/or necrosis. Recently, it was reported that the expression of the major executor caspase-6 was upregulated in a p53-dependent manner in neurons and skeletal muscle cells expressing mutant HTT protein.[93] The authors proposed that the presence of the mutant HTT protein stabilized p53 or enhanced its pro-apoptotic effects. Expression of mutant human SOD1 in transgenic mice modelling MND resulted in upregulation of p53 and factors of receptor-mediated apoptosis, such as Fas, and increased the rates of motor neuron death by necrosis.[94] It is possible that locally acting factors may decrease the threshold for apoptosis in some cell populations but not others, even populations of cells of the same type.

Cancer and LONDD: extremes of a wide spectrum?

A somewhat unusual relationship has been noted between common LONDDs and cancer. Cancer and neurodegenerative disease are both quite common in advanced age, but, nevertheless, multiple reports state that there was significant inverse comorbidity between them.[95–98] The initial association was made for dementia and cancer, [95] but since AD is the most common cause for dementia in the elderly and a significant part of patients with PD also eventually develop dementia, the association is likely to be valid after stratification by disease. Later, the existence of inverse comorbidity with cancer was confirmed for PD and HD.[99,100]

It is possible that common mechanism/s may be involved in the pathogenesis of LONDD and cancer. These mechanisms are probably subject to modulation by other genetic and/or environmental factors so that the potential outcomes would rarely coincide in the same patient. Accumulation of errors in DNA is now known to be a significant part of the mechanism of carcinogenesis. The mechanisms for identification and repair of damage in DNA and the associated mechanism for maintenance of genomic integrity are the key mechanisms for safeguarding cells from cancerous transformation (reviewed in [101]). Cell and tissue aging (and the associated late-onset degenerative diseases) is currently believed to be directly related to the age-related decrease of proliferative potential of the adult stem cell niches, which is also closely related to accumulation of unrepaired damage and replicative errors in DNA.[101] Whether after many decades of daily genotoxic barrage the outcome would be degenerative disease or cancer, or both, is a product of the complex interaction between genetic factors, environmental factors and, for lack of a better word, pure chance.

Cancer and LONDD are both common after the age of 60 (and tend to become more common as age advances) and are both among the 10 leading causes of mortality in the elderly. For both, peak prevalence is in the seventh to the eighth decade, then, in the ‘oldest old’ they become less common. Is it possible that these two represent the extremes of a spectrum? Studies show that the relationship between risk of LONDD (best studied for AD) and cancer is bidirectional, i.e. AD patients are at decreased risk of developing some of the common types of cancer than the general population and, at the same time, cancer survivors are at significantly (20%–50%) decreased risk of developing AD. A similar association of PD with decreased risk of almost all common cancers, including those that are strongly related to smoking (lung, colorectal and bladder cancer), was reported in 2007.[102,103] There were two notable exceptions: melanoma and breast carcinoma that were actually more common in patients with PD than in age-matched controls. Similarly, in individuals with Down syndrome aged over 30 years, the risk of leukaemia (specifically, acute myeloid leukaemia, but also acute lymphoblast leukaemia (ALL)) is significantly elevated, whereas the risk of most solid tumours (including breast cancer) is actually lower than the population risk.[104] Thus, the risk for most common cancers is decreased in patients with AD and PD, but is increased for several specific cancers that may be different for patients with AD and PD. What is particularly interesting, however, is that virtually all cancers that are common in AD and PD patients are cancers for which the existence of cancer stem cells has been suspected and/or confirmed.[105–107]

The decreased cancer risk seen in LONDD may be explained by the capacity for more efficient disposal of cells that have sustained potentially carcinogenic damage. In a genetic background conferring increased propensity to eliminate cells with damaged DNA, subtle deficiencies of BER and/or randomly occurring deletions of mitochondrial DNA increasing the levels of oxidative damage (see later) may be associated with increased risk of degenerative disease and, especially, neurodegenerative disease. The risk would increase with age until it reaches a maximum in a specific age group (for LONDD, 80–85 years). By that time the majority of the individuals carrying the combined genotype of deficiency of repair/proneness to apoptosis would have developed the associated disease. Beyond this age, it is likely that those that still are neurologically intact have remained so either because they had inherited a genetic makeup consistent with very efficient DNA repair (in which case it is that they have remained in good overall health despite the advanced age) or because they have inherited a set of gene variants conferring ‘pro-repair’ rather than ‘pro-apoptotic’ tendency (in which case one could expect that they may have or have had some type of cancer). In many common types of cancer, subtly decreased capacity for repair of DNA damage may be associated with better chances of success of genotoxic therapy because of increased chances that cancer cells would rapidly accumulate damage and die or may arrest their rapid cycling at least temporarily to attempt to repair the damage. Thus, individuals with slightly decreased repair capacity may be at increased risk of cancer but may have better outcomes after anticancer treatments and, respectively, may live to an old age after cancer has been eliminated.[108] Differentiated neurons in these individuals may suffer increased burden of damage, but since they manage to maintain the integrity of the important regions of their DNA, it is likely that neural tissue would be spared from premature death due to accumulation of damage and carcinogenic transformation would be highly unlikely. Thus, some of the very healthy ‘oldest old’ and the ‘oldest old’ that have survived common cancers may be expected to remain free of neurodegenerative disease until the end of their lives.

In cancers associated with permanent blockade of the differentiation pathway (as most cancers originating from cancer stem cells are), however, this mechanism would probably not work as well. Cells that have been arrested early in the course of their differentiation are likely to have retained their high proliferative potential by abrogating some of the regulatory mechanisms that would normally divert the cell towards the apoptotic pathway in case it does not fulfil the requirements for the next stage of differentiation. This would specifically be the case for cells that are normally subjected to repeated selection procedures during their differentiation (differentiating blood cells) and cells that are programmed to divide rapidly in the course of their differentiation (blood cells and epithelial cells). Thus, the risk of most cancers (requiring increased levels of genotoxic stress as well as increased tolerance of errors in DNA) would be lower in patients with LONDD, with the exception of haematological cancer and some tumours of epithelial origin that have found a way to override pro-apoptotic signalling. However, the mechanism is probably much more complex.

Notably, no inverse association was found between the risk of cancer and vascular cognitive impairment (VCI).[96] VCI is an umbrella term referring to cognitive decline after single or multiple strokes and/or clinically silent brain microhaemorrhages. AD and VCI may co-exist in the same patient (mixed dementia). The role of acute and chronic genotoxic stress in neuronal death in stroke has been well studied,[109,110] and the individual capacity for management of genomic integrity has been shown to be a significant factor in the establishment of the risk of stroke and its potential outcomes.[111–113] Cerebral amyloidosis may play significant role in vascular disease, stroke and VCI, although the type of amyloid that is deposited in the extracellular space of the brain parenchyma and the amyloid that is deposited in the vessel wall are different.[114] Atherosclerotic plaque, however, is also a major factor in the pathogenesis of vascular disease and stroke. The inflammatory response of the endothelium in the area affected by atherosclerotic plaque may also result in breach of the integrity of the vascular wall. Thus, neuronal pathology is, most likely, only part of the pathogenesis of vascular dementia. It is possible that cell death due to the combination of increased genotoxic stress (because of genetic and/or environmental factors) and increased propensity to divert cells towards the apoptotic pathway would account only for a proportion of cases of VCI.

Individual capacity for DNA repair and maintenance of genomic integrity: a potential tool in the assessment of risk of late-onset disease

The role of oxidative stress in the pathogenesis of multifactorial disease has been noted since the early 1990s, when it was demonstrated that increased levels of oxidative stress were a molecular hallmark of insulin-resistant diabetes.[115] At the time, the causes for increased genotoxic stress were believed to be environmental rather than genetic, as human genetic diseases due to deficiency of BER were not known. The latter was usually explained as a consequence of the severe phenotypes produced by deficiencies of BER that inevitably caused early intrauterine death. Not until 2006 was a direct causal relationship discovered between carriership of polymorphic variants of genes coding for major enzymatic activities of BER conferring a subtle decrease in activity (specifically, the glycosylases Ogg1 and Neil1) and the risk of development of metabolic syndrome and/or diabetes type 2 in mice.[116] Later, the association was confirmed for humans.[117] In 2007, it was shown that subtly decreased capacity for repair of oxidative lesions in DNA due to carriership of polymorphic variants of Ogg1 could modulate the effects of the extended CAG repeat in the Hdh gene in mouse models of HD.[59] Profound impairment of BER may indeed be incompatible with life, but mild BER deficiencies are apparently quite common and may play a role in the susceptibility to common late-onset disease.

It is now known that the capacity to detect and repair DNA damage may vary significantly even between clinically healthy individuals due to genetic variance in the efficiency of the mechanisms for detection and repair of DNA damage.[118] The role of the two principal components of individual repair capacity (the capacity to identify and repair damage and the capacity to assess genomic integrity) has been already studied at length in relation to the risk of cancer and the potential outcomes after anticancer treatment. Subtle deficiencies in different types of DNA repair have been strongly implicated in individual susceptibility to cancer as well as in eligibility for various genotoxic therapies (reviewed in [108,119]). Nevertheless, the role of individual repair capacity in susceptibility to human diseases other than cancer is still largely unexplored. Not all of the genetic factors that may play a role in the establishment of individual risk of late-onset disease have been thoroughly studied. The results reported so far show a fair amount of variation between populations. Despite the large amount of information already accumulated in the field of assessment of the individual risk of developing common late-onset diseases, it is difficult to make a reliable risk assessment at present, even in individuals with positive family history. It is believed, nevertheless, that the unravelling of the genetic bases of late-onset disease is, at present, a matter of time and effort. The more we know about the mechanisms of multifactorial disease, the sooner we could expect that the information may be put to practical use in the assessment of the risk of various age-dependent diseases, and, potentially, for their prevention and/or treatment.

Oxidation is the predominant type of damage in differentiated neurons and, respectively, BER is the dominant type of repair, with NER being selectively employed only in the transcribed regions of the genome. Thus, the activity of the cellular machinery of BER is crucially important for the normal functioning of neurons. Respectively, deficiency of BER may increase the risk of development of neurodegenerative disease. So far, results from studies in mouse and rat models support this hypothesis. A recent study carried out in transgenic mice carrying mutant copies of the APP, PSEN1 and MAPT human genes (all implicated in the pathogenesis of familial AD) reported that deficiency of DNA polymerase beta accelerated the onset of cognitive decline and exacerbated the AD phenotype.[120] Results obtained in rodent models, however, do not always translate directly to primates and humans. Normal human neurons have been shown to be relatively deficient in molecules of long-patch BER (FEN-1, polymerases delta and epsilon, ligase I and its auxiliary factor PCNA) and rich in proteins of short-patch BER and NER (XRCC1 and ligase III), indicating that single-nucleotide BER might be preferred over long-patch BER in human neurons, probably because of the lower relative risk of introduction of errors in template DNA due to inaccurate copying.[121] So far, the results of association studies between carrierships of variant alleles modulating the capacity for BER and the risk of LONDD in man are conflicting. At least two groups have failed to find association between carrierships of polymorphisms in the hOGG1, APE1 and XRCC1 genes and the risk of sporadic AD.[122,123] Polymorphisms in two of these three genes (APE1 Asp148Glu and XRCC1 Arg399Gln), however, together with a common polymorphism in a gene coding for a component of the machinery for repair of strand breaks by homologous recombination (XRCC3 Thr241Met) have been found to be associated with increased genetic risk of PD.[124] A weak association has also been found between carrierships of the Cys allele of the common Ser326Cys (rs1052133) polymorphism in the human OGG1 gene (the Cys326 allele being associated with lower repair activity than Ser326 allele) and the risk of sporadic ALS.[125,126] No association between the APE1 polymorphism Asp148Glu and the effect on the risk of sporadic ALS was identified in a later study by the same authors.[127] Apparently, different polymorphisms conferred unequal risks of different neurodegenerative diseases.

Notably, variant alleles of some of the polymorphisms that were identified to have a modulating effect on the risk of LONDD have previously been demonstrated to play a role in the genetic risk of various cancers – almost invariably, decreasing the risk rather than increasing it. The Cys/Cys genotype of the hOGG1 Ser326Cys (implicated as risk modulator in PD) has been previously associated with decreased risk of prostate cancer [128] but, at the same time, with increased risk of senile cataract (another age-related degenerative disease).[129] Similarly, the Glu allele of the APE1 Asp148Glu polymorphism was shown to have a minor protective effect against prostate cancer in some populations,[130] but is, at the same time, one of the polymorphisms identified to increase the risk of PD in the study of Gencer et al. [124] cited above. This may be another example of the inverse relationship between age-dependent degenerative disease and cancer, brought about by a lifelong compromise between accuracy of identification and repair of damage and the propensity to route damaged cells to apoptosis. Interestingly, children carriers of Cys/Cys genotypes by the Ser326Cys polymorphism of hOGG1 were identified to be at increased risk of development of ALL than carriers of at least one Ser allele, [131] supporting the hypothesis that the risk of LONDD was associated with decreased risk of most cancers except for those that originated from cells with differentiation blocks (as is ALL).

Many of the factors that play a role in the assessment of the integrity of the genome have been identified, although details about their precise function in this assessment have been identified for only a few of them – mainly, p53 and ATM.[132–134] The TP53 gene codes for the p53 protein, which is the major controller of cell cycle progression and cell death. The gene is conserved and only a few polymorphisms have been described so far. One of these polymorphisms, the single-nucleotide polymorphism rs1042522 (Pro72Arg) in the TP53 gene [135] has direct effect on the assessment of whether to repair genomic damage or route the damaged cell towards apoptosis. Both allelic variants of the TP53 Pro72Arg polymorphism are essentially wild-type, although the 72Pro variant is a stronger inducer of cell cycle arrest in the presence of damage than the 72Arg allele, whereas 72Arg is a stronger inducer of apoptosis in the presence of damage than 72Pro.[132] Thus, cells carrying the ‘pro-repair’ 72Pro allele/s would, rather, try to repair the damage, whereas cells carrying the ‘pro-apoptotic’ 72Arg allele would, rather, divert the dam-aged cell to the programmed cell death pathway. It is possible that carriership of the Arg allele of the TP53 Pro72Arg polymorphism and potentially of other polymorphisms in the genes coding for proteins directly involved in the assessment of the integrity of the genome may modulate the risk of late-onset disease and the potential outcomes. The studies about the role of TP53 polymorphisms in LONDD are still very few, although it has been shown that p53-dependent pathways that are often upregulated in colorectal and lung cancer (related to expression of cancer-specific isoforms of p53) may be downregulated in AD and PD (probably in an effort to compensate for the mass cell death).[136] To date, association studies have failed to elicit a direct link between carrierships of the one or the other allele variant of Pro72Arg and LONDD, but as there have been only two studies in the field so far [137,138], it is still very early to draw conclusions. Most components of individual repair capacity have only a minor contribution to the risk when taken separately, the risk being modified and modulated by the plethora of other genetic and environmental factors. Nevertheless, unrelated studies show that the ‘pro-repair’ Pro/Pro homozygous genotype was more common in healthy nonagenarians and centenarians than in the younger old and that its carriership was associated with increased survival in all age groups, including survival after cancer.[139,140] As mentioned above, it was reported that cancer survivors had an inherently lower risk of development of LONDD (20%–50%, according to different studies). It is possible that normal or near-normal capacity for DNA repair (conferred by genetic factors) coupled with other genetic factors associated with a ‘pro-repair’ propensity may increase the risk of cancer but, at the same time, decrease the risk of age-dependent damage of neural tissue. It could be expected that antioxidant therapy (an environmental factor) may complement the genetic background to decrease the risk of both cancer and LONDD by decreasing the level of ROS and, respectively, the amount of genotoxic damage.

Studies of the associations between LONDD and deficient DNA repair/maintenance of genomic integrity are still primarily phenomenological, as are most studies in the field of individual repair capacity. With the current rapid rate of accumulation of data in the field, it may be expected that the potential relationship between repair capacity/capacity for assessment of integrity of the genome and LONDD may become clearer in the near future. Extensive studies into the role of DNA damage identification and repair and assessment of genomic integrity may be needed in order to elucidate the role of individual repair capacity in the risk of development of late-onset disease. Knowledge about personal genetic status with regard to risk of neurodegenerative disease may potentially assist in the selection of adequate pharmacological interventions and/or lifestyle modifications (e.g. antioxidant therapies) to delay or prevent development of neurodegenerative disease.

Conclusions

Differentiated CNS neurons accumulate significant amounts of genotoxic damage (specifically, oxidative damage) that increases with age. Neurons selectively focus the larger part of their repair capacity to the transcribed regions of their DNA, suppressing damage identification and repair in the untranscribed regions. As a result, not all genomic damages are included in the overall assessment of the scale and the scope of damage. Thus, the overall threshold of damage beyond which the cell is assessed as irreparably damaged may be higher in normal neurons than in other types of cells in order to extend the life and preserve the functionality of the neuron. Under conditions of subtly decreased capacity for repair of oxidative DNA damage (e.g. due to carriership of variant alleles of genes coding for major players in DNA damage identification and/or repair), the damage burden may become too much for the cell and may be assessed as irreparable so that the cell may be sent along the programmed cell death pathway. Genetic propensity for routing damaged cells to apoptosis instead of attempting repair of damage may increase the risk that neurons that have sustained moderate degree of damage may be assessed as severely damaged and may, respectively, be routed towards apoptosis. Normally, CNS neurons are rarely replaced (if ever), as the adult neural progenitor cell niches in the CNS are sparse and their proliferative capacity is limited. Overstimulation of the neural stem cell niche due to increased loss of neurons with irreparably damaged DNA may eventually deplete its proliferative potential and may trigger abnormal re-entry of differentiated neurons in the cell cycle. The latter would also, eventually, result in neuronal death. We hypothesize that the pathogenesis of LONDD may be at least partly associated with life-long decreased capacity to manage DNA damage leading to accumulation of unrepaired damage and subsequent aberrant induction of apoptotic mechanisms; followed by overstimulation of the brain neural stem cell niche, depleting its proliferative potential. Initially, the rate of loss would slightly outweigh the rate of production of new neurons (onset of neurodegeneration), then, as age advances and the efficiency of the mechanisms that protect neurons from damage declines, more and more neurons would die and no new functional neurons would be able to be produced (end-stage neurodegenerative disease). The end stage may be reached sooner in some individuals and later in others, depending of the rate of neuronal death and the rate of their replacement (and, respectively, the rate of depletion of the stem cell niche). One could suppose that this end stage may eventually be reached in normal aging, but only after a very long time (after the ninth decade of life). The hypothesis that LONDD may stem from a combination of genetic traits predisposing to decreased capacity to repair damage and increased propensity to route cells that have sustained damage to apoptosis makes an interesting contrast with the genetic risk of other common late-onset disease, namely, cancer. It is presently believed that the genetic bases of cancer involve decreased capacity for detection and repair of errors in DNA and/or decreased capacity to route cells that have sustained too much damage to apoptosis. Thus, LONDD and cancer may represent the two extremes of the wide spectrum of the individual capacity for repair of genotoxic damage and maintenance of genomic integrity. While genetic factors are, at present, difficult to control, many of the potential environmental factors are controllable. Further research into the possibilities for targeted modulation of genetic risks of late-onset disease is needed in order to delay development and/or ameliorate the course of disease of advanced age. Meanwhile, individuals at risk may benefit from antioxidant therapies, pharmacological as well as dietary, in order to decrease the risk and/or delay the onset of neurodegenerative disease.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research was supported by the National Science Fund, Ministry of Education and Science of Republic of Bulgaria [grant number DFNI-B01/2].